Table 1 Hyperparameters for the developed models
From: Modelling customers credit card behaviour using bidirectional LSTM neural networks
Model task | Parameter value |
|---|---|
Number of transactional features | 3 |
Number of additional features | 8 |
Number of bidirectional LSTM layers | 2 |
Number of cells in each LSTM layer | 4 |
Number of attention layers | 1 |
Activation function of hidden LSTM layers | Hyperbolic tangent |
Number of fully connected layers (dense) | 1 |
Activation function for all dense layers except the output one | Sigmoid |
Activation function of the output layer | Sigmoid |
Loss function | Binary cross-entropy |
Optimizer | Adam |